Blueprints AI is preparing to show a new version of its construction-document software at TechCrunch Disrupt in San Francisco this week. The company’s launch page lists October 13 through 15 and promises drawings produced in minutes. Architects and developers are being offered a faster way to assemble documents that can otherwise require coordination among several specialists. Its public contracts also describe another potential source of revenue: payments for placing products inside those documents.
Under a section headed “Promotions and Product Placement,” Blueprints AI says it may promote products in generated designs and receive commissions or other payments. Users are asked to accept that the arrangement creates no conflict among stakeholders. The clause establishes permission; the documents examined for this article do not identify a paying manufacturer, a particular placement or money collected. It nevertheless gives design firms a concrete issue to resolve before a generated specification becomes part of a client’s project.
The range of documents makes that provision consequential. Blueprints AI describes output spanning architectural, structural, mechanical, electrical and plumbing drawings, plus energy, fire-safety and interior sheets. Mechanical sheets include equipment schedules; electrical sheets include panels and lighting; interior sheets include finishes. These are places where a design can identify what contractors are expected to install. The company also offers civil and landscape generation on enterprise plans.
Blueprints AI says changes requested in ordinary language update a coordinated set of drawings. Customers can start with sketches, photographs, existing plans, point-cloud measurements or text, then review and mark up the output. The company requires licensed professional review before permit submission and a stamp where applicable. Those review requirements put architects and engineers in a position to assess product choices along with dimensions, calculations and code references. SnowRock has not tested the system or examined a customer’s completed permit set.
Architects have their own rules about supplier payments
California’s Architects Information Guide states that architects cannot solicit or accept payments, rebates, refunds or commissions from material or equipment suppliers in exchange for specifying their products. It also sets written-disclosure requirements for compensation from multiple parties and for financial interests substantial enough to influence professional judgment. Those provisions address architects’ conduct. Establishing how they apply to a particular software arrangement would require knowing who received money, what services they performed and how products were selected.
A vendor’s contractual statement about conflicts therefore needs to be considered alongside a design firm’s own circumstances. An architect might be paying for software while a separate company receives promotional revenue. The public clause alone cannot establish the architect’s financial interests or resolve a licensing question. What a firm can examine before purchase is whether it can identify paid selections, replace them and explain its final choices to a client. The company’s published placement provision does not describe those controls.
The American Institute of Architects’ current ethics code, dated June 9, has a separate disclosure rule for its members. When other responsibilities or personal interests could affect professional judgment, people relying on that judgment must consent after full disclosure. Its commentary identifies clients, owners, employers and contractors among those who may be affected. This is a membership rule, with an institute disciplinary process, rather than a government finding about Blueprints AI.
The same code requires members to maintain responsible control over drawings and specifications they sign or seal. AIA describes that as the knowledge and supervision ordinarily expected in professional practice. Members also owe clients confidentiality. Taken together, these duties cover several decisions a firm makes when introducing AI: what information it uploads, how it checks proposed specifications and who takes responsibility for finished documents. They continue to apply when another company supplies the drafting software.
California’s stamp guidance explains the public purpose of an architect’s signature and seal: they identify responsibility for plans, specifications and other professional documents. The stamp carries an architect’s licensed name, license number and renewal information. A building department can identify a professional answerable for a submission even when several people or software systems helped prepare it. Choosing a production tool changes how documents are made while leaving that named professional visible on them.
Blueprints AI is building professional participation into its service through a directory of architects and engineers. It says it checks licenses against state-board records before listing practitioners and lets customers contact them directly. Participants can advise on design, comment inside the application and revise work as an agent drafts. Some engagements involve serving as architect or engineer of record, the professional responsible for the relevant discipline.
That arrangement could bring review into a project before a completed set reaches a professional’s desk. The company says its application keeps comments, markups and project history together, and that engagements involving a seal must meet state requirements for responsible control or charge. Public descriptions do not show how much time participating professionals spend reviewing generated work. They do, however, describe an ongoing role more substantial than adding a signature to a finished drawing.
Customer drawings help tailor the software
Blueprints AI also depends on records supplied by its customers. Its onboarding explanation describes roughly two weeks of adapting a model to a firm’s past projects, usual building types, company standards and comments from local permitting authorities. Firms upload title blocks, sheet layouts, detail libraries and naming conventions. Those materials help software reproduce a practice’s established drawing style and respond to requirements that vary by jurisdiction.
Comments from a plan reviewer provide a different kind of information from a completed drawing. They identify issues an authority asked a design team to address. Blueprints AI explicitly includes those comments in its account of how it tailors its service. Whether a particular archive is useful will depend on how clearly a firm can connect comments to the submitted version, its revisions and the requirements in force at that time. That is an inference from the described process, not a measured result reported by the company.
California offers a current example of why dates matter. The state’s 2025 Energy Code applies to buildings with permit applications submitted on or after January 1, 2026. It expands heat-pump requirements for new residential construction, encourages electric readiness and strengthens ventilation standards. A plan approved under an earlier edition belongs to a different regulatory period. A system using past work to prepare a new application has to account for that change.
The California Energy Commission provides compliance manuals, approved performance-modeling software, forms and reference appendices containing weather data and information on building assemblies. These materials serve different purposes: a drawing shows proposed construction, while a compliance calculation assesses it against applicable requirements. Blueprints AI’s promise to produce energy-compliance sheets can be evaluated against those specific outputs and methods. Merely reproducing a firm’s sheet layout would leave the substantive calculation to be checked.
Data use introduces a second contractual question. The same terms permit Blueprints AI to use customer data to train and improve its models. Another section says uploaded files, project documents and sensitive personal details remain confidential and are never published. Training permission and a promise against publication concern different activities. The terms also say generation times are illustrative and vary with project complexity, software version and plan. Firms buying the service therefore need both a data-use agreement and a realistic allowance for professional review.
The company’s pricing page makes a more specific statement: project data is used only for the customer’s company and is never shared with other users. It advertises an option to run software on a customer’s own premises for enterprise engagements. Its public material does not explain the technical separation between firm-specific adaptation and general model improvement. That detail would determine how the broad training permission operates for a customer’s drawings.
Blueprints AI’s privacy policy, revised July 26, says uploaded sample projects and drawings are limited to staff and service providers who need them to handle requests and deliver services. It permits use of anonymized form answers, such as descriptions of a customer’s needs, in marketing and product development. Those statements distinguish project files from descriptions collected during sales inquiries, a distinction relevant when firms supply examples while arranging a demonstration.
Architectural records can also contain several kinds of protected work. The U.S. Copyright Office’s registration guidance distinguishes a building’s architectural design from technical drawings depicting it. Separate registrations cover those works. Protection for architectural designs can include overall form and arrangements of spaces, while standard components and purely functional features receive different treatment. Deciding which project records a firm may supply to a software provider requires attention to authorship and existing agreements, beyond possession of a PDF.
AIA’s AI toolkit for firms addresses that contracting problem directly. It tells architects to examine owner agreements for restrictions on training with project data, required disclosures, ownership provisions and limits on AI-generated deliverables. A firm that accepts such conditions from an owner also needs to pass relevant requirements to consultants. The guidance brings engineers, specification writers and other contributors into the same discussion, because records may cross several businesses before a permit application is assembled.
The toolkit also calls for review suited to the consequences of a task and documentation of AI use in project records. For a service that generates coordinated specifications, that would support examining product selections, calculation inputs and revisions alongside the final sheets. Blueprints AI lists booth BW1 at Moscone West for October 13 through 15 and says V2 will become public during its stage presentation. Details of that release remain withheld on its launch page.
